def test_sampler_good_starting_point(self):
"""Test running sampler with valid starting point"""
input_path = os.path.join(TEST_DIR, '2d.txt')
constraint = Constraint(input_path)
sampler = Sampler(constraint)
results = sampler.sample(100)
self.assertEqual(len(results), 100)
self.assertTrue(all(constraint.apply(pt) for pt in results))
def test_sampler_bad_starting_point(self):
"""Test running sampler with invalid starting point, where
the sampler has to find an edge first"""
input_path = os.path.join(TEST_DIR, '2d-badstart.txt')
constraint = Constraint(input_path)
sampler = Sampler(constraint)
results = sampler.sample(100)
self.assertEqual(len(results), 100)
self.assertTrue(all(constraint.apply(pt) for pt in results))
def test_apply_constraint_func(self):
"""Test applying constraint function to a point to retrieve the value
of the constraint function.
"""
input_path = os.path.join(TEST_DIR, '2d.txt')
constraint = Constraint(input_path)
self.assertAlmostEqual(
Sampler._apply_constraint_func(
constraint.get_constraint_funcs()[1], [0, 1]), -0.2)
self.assertAlmostEqual(
Sampler._apply_constraint_func(
constraint.get_constraint_funcs()[1], [0, 0.5]), 0.3)
def main(input_file, output_file, n_results):
"""Runs the sampling algorithm on the problem defined in the
constraint input file INPUT_FILE and outputs N_RESULTS number
of sampled points to OUTPUT_FILE.
"""
constraint = Constraint(input_file)
sampler = Sampler(constraint)
samples = sampler.sample(int(n_results))
with open(output_file, 'w') as f:
for point in samples:
vector = " ".join(str(v) for v in point)
f.write(vector + '\n')
def test_step_to_next_point(self):
"""Test stepping function"""
current_pt = np.array([0.5, 0.5])
step_vector = np.array([0, 1])
step_vector = step_vector / np.linalg.norm(step_vector)
mag = 0.2
next_pt = Sampler._get_new_point_after_step(current_pt, step_vector,
mag)
self.assertTrue(np.allclose(next_pt, [0.5, 0.7]))
def test_apply_constraint_to_step_candidate(self):
"""Test applying constraint function to a point that is stepped to
in order to retrieve the value of the constraint function."""
input_path = os.path.join(TEST_DIR, '2d.txt')
constraint = Constraint(input_path)
current_pt = np.array([0.5, 0.5])
step_vector = np.array([0, 1])
step_vector = step_vector / np.linalg.norm(step_vector)
mag = 0.2
value = Sampler._apply_constraint_to_step_candidate(
constraint.get_constraint_funcs()[1], current_pt, step_vector, mag)
self.assertAlmostEqual(value, 0.1)
def test_is_valid_point(self):
"""Test if points lie in the unit hypercube and satisfy constraints"""
input_path = os.path.join(TEST_DIR, '2d.txt')
constraint = Constraint(input_path)
sampler = Sampler(constraint)
# Satisfies constraints
self.assertTrue(sampler._is_valid_point(np.array([0.5, 0.5])))
# Does not satisfy constraints
self.assertFalse(sampler._is_valid_point(np.array([0.1, 0.1])))
# Outside cube
self.assertFalse(sampler._is_valid_point(np.array([-0.1, 0.1])))
self.assertFalse(sampler._is_valid_point(np.array([0.1, -0.1])))
self.assertFalse(sampler._is_valid_point(np.array([1.1, 0.1])))
self.assertFalse(sampler._is_valid_point(np.array([0.1, 1.1])))